Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866911734629924864 |
|---|---|
| author | Park, Junyoung Ju, Seongyong Park, Sunghwan Lee, Jaewoo |
| author_facet | Park, Junyoung Ju, Seongyong Park, Sunghwan Lee, Jaewoo |
| contents | As Large Language Models evolve for user convenience, vulnerability to jailbreak attacks continues to be reported despite ongoing efforts in safety training. Traditional jailbreak techniques typically focus on a single prompt injection, neglecting the models' ability to remember the flow of conversation and the user's instructions. In this paper, we propose Persona Attack, a memory injection based jailbreak method that manipulates the model's context window through a step by step approach. Experimental results from applying Persona Attack to several widely used LLMs reveal that, as injections accumulate in memory, models increasingly prioritize these instructions over their internal safety alignment mechanisms. Furthermore, our experiments empirically demonstrate that the attack success rate varies not only according to the memory implementation of the model, but also combinations of instructions and can reach 95% under specific instruction configurations. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2606_00150 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models Park, Junyoung Ju, Seongyong Park, Sunghwan Lee, Jaewoo Cryptography and Security Artificial Intelligence As Large Language Models evolve for user convenience, vulnerability to jailbreak attacks continues to be reported despite ongoing efforts in safety training. Traditional jailbreak techniques typically focus on a single prompt injection, neglecting the models' ability to remember the flow of conversation and the user's instructions. In this paper, we propose Persona Attack, a memory injection based jailbreak method that manipulates the model's context window through a step by step approach. Experimental results from applying Persona Attack to several widely used LLMs reveal that, as injections accumulate in memory, models increasingly prioritize these instructions over their internal safety alignment mechanisms. Furthermore, our experiments empirically demonstrate that the attack success rate varies not only according to the memory implementation of the model, but also combinations of instructions and can reach 95% under specific instruction configurations. |
| title | Persona Attack: Incremental Memory Injection Jailbreak Attack against Large Language Models |
| topic | Cryptography and Security Artificial Intelligence |
| url | https://arxiv.org/abs/2606.00150 |